CAE Answer

Answers for engineers who read FEA results

Short, practical notes on stress, stiffness, energy and load paths, and on checking a model before you trust its numbers.

Mohr's circle for pure shear A stress element under pure shear of 100 MPa rotates 45 degrees. On Mohr's circle the matching diameter rotates 90 degrees, from the shear axis to the normal-stress axis, where it meets sigma 1 equals plus 100 MPa tension and sigma 3 equals minus 100 MPa compression. σ τ +100 −100 σ1 σ3

Pure shear, τ = 100 MPa, with no normal stress at all. Turn the element 45° and the shear disappears: σ1 = +100 MPa tension, σ3 = −100 MPa compression. On Mohr's circle the same turn is 90°.

Key ideas

The mechanics behind most questions about a fringe plot or a force curve.

Area moment of inertia

Ixx = ∫ y² dA

Material far from the bending axis counts with its distance squared. A standing I-beam has a large Ixx, so it is stiff vertically, and a small Iyy, so it bends easily sideways. Same beam, same material, different axis.

Bending stiffness is E·I. A steel and an aluminium beam of the same shape have the same I but a different EI.

Not the same as mass moment of inertia (kg·m²), which governs how hard a body is to spin. Try it with a plastic ruler: bend it across its flat face, then across its edge.

Principal stress

σ1,2 = (σx + σy)/2 ± √[ ((σx − σy)/2)² + τxy² ]

The six stress components at a point depend on the axes you choose. Rotate the axes until every shear stress is zero: the normal stresses left are the principal stresses. σ1 is the largest tension on any plane through the point, and σ3 the largest compression.

There are always three, on mutually perpendicular planes, because the stress tensor is a symmetric 3×3 matrix. The principal stresses are its eigenvalues and the principal directions its eigenvectors.

Von Mises or principal stress?

Von Mises folds all six components into one positive number to compare with the yield strength from a tensile test. It suits ductile metals: steel and aluminium.

It has no sign and no direction, so it cannot tell tension from compression, and a brittle crack can hide on a von Mises fringe. For glass, castings, lamp lenses and plastic clips, judge with σ1. Cracks open perpendicular to its direction.

Once most of a part has yielded in an impact, effective plastic strain says more than either.

Energy absorption

E = ∫ F ds ≈ Favg × stroke

Energy is the area under the force–displacement curve, not its peak. A structure that is too stiff reaches a high force with little crush and passes the load downstream. One that is too soft bottoms out before it has absorbed enough.

The aim is a flat, sustained crush force over a long stroke. At a comparable force level, the longer available stroke wins, which is why standoff distance is often the best way to rank two designs before running any FEA.

Compare designs of different mass with specific energy absorption, SEA = E / mass. Judge how flat the curve is with crush force efficiency, CFE = mean force / peak force; closer to 1 is better. A higher I tunes how a section folds; it does not raise its capacity.

Stress is not force

Stress is force per area. A thick bracket can carry a large force at low stress, and a thin flange can show high stress while carrying little.

Principal-stress vectors show where a load path runs. A section force (*DATABASE_CROSS_SECTION, written to secforc) tells you how much force goes through it. It is the free-body-diagram cut from a mechanics course, done inside the model.

A useful check: put a cut on each branch of the path and confirm the branch forces add up to the applied load.

Load paths as stress trajectories

Join the principal directions point to point and you get stress trajectories. Those curves are the load path. In a simply supported beam with a centre load, compression arches from the load to the supports while tension hangs underneath like a cable, and the two families always cross at 90°. The part behaves like a hidden truss.

Adding material along a trajectory stiffens the path. A bead or notch across a compression trajectory makes the member buckle early instead of passing the load through.

Before you trust a result

A broken explicit model still produces believable numbers. Check these on every run.

CheckWhereRule of thumb
Energy balanceglstatTotal energy stays roughly flat; energy ratio ≈ 1.0
Hourglass energyglstat, matsumUnder 10% of internal energy, globally and for each part
Contact energyglstat, sleoutSmall; clearly negative means a contact problem
Added massglstat, d3hspSmall compared with the parts in the load path
Initial penetrationsmessag, d3hspNone in the region of interest
Time stepd3hspNot controlled by one tiny or distorted element

These are common industry rules of thumb for LS-DYNA, not values from a standard. Where your team has its own acceptance criteria, use those. Check the boundary conditions too: supports and constraints should match the physical test.

Which output answers which question

Finding the part that causes a peak means overlaying the right curves.

rcforc: contact forces
Which contact pair carries the load, and when. Plot each one against the response curve; the contact whose force rises with the spike is the cause.
secforc: section forces
How much force and moment pass through a cut plane, such as a bracket, hinge or beam.
nodout: displacements
Travel against the available clearance, to show when two parts meet.
Effective plastic strain
What yields, and where. More useful than von Mises once most of the part has yielded.
Principal-stress vectors
Load-path direction and brittle-crack risk. Pair them with secforc for magnitudes.
d3plot with a section cut
Step frame by frame through the moment of the peak. This is what convinces the part engineer.

Traps worth knowing

Each of these has produced a wrong conclusion from a correct-looking plot.

  • Shell surfaces. In bending, the top surface can be in tension while the bottom is in compression. A mid-surface plot can show almost nothing while both outer surfaces are near failure. Use the outer surfaces for fracture checks and the mid-surface for membrane load paths.
  • σ3 = 0 on a shell. Shells are in plane stress, so the through-thickness stress is zero and it gets sorted into one of the three principal slots. A 3rd-principal fringe of zero everywhere means the panel is pulled in both in-plane directions, not that it carries no compression.
  • Failure that isn't modelled. A clip, bracket or glass panel that should break but has no failure definition becomes a false hard point.
  • Strain-based failure. Many LS-DYNA failure criteria for plastics and glass use principal strain, not stress. Read the material card before judging a fringe against a limit.
  • The load path changes with time. Parts make contact, buckle or crack during an impact, so the path early in the event can differ completely from the one at the peak. Step through the frames.
  • CATIA Measure Inertia. 2D mode on a section face gives area moments in m⁴. 3D mode on the same face gives a mass moment of inertia in g·mm², computed with a surface density. Use 2D, read the values through the centroid (Inertia / G, never / O), and select only the faces of the member you are comparing.

Learn in layers

For a practising engineer. Stop at the layer you need.

  1. Mechanics of materials

    Mechanics of Materials by R.C. Hibbeler, or Gere & Goodno; the content is the same. Read the chapters on stress, mechanical properties and bending, and the von Mises part of failure theories. Skip torsion, shear flow, deflection, buckling formulas and energy methods.

    You can skip the book if you can answer all three:

    1. Why does a thin plate bend easily, but a ribbed or hat-section part does not?
    2. What is the difference between yield and failure? What does a plastic strain of 0.15 mean?
    3. What does a section-force value physically represent?

    If one answer is fuzzy, read only that chapter.

  2. FEA for practitioners

    • Building Better Products with Finite Element Analysis, Adams & Askenazi
    • Finite Element Analysis for Design Engineers, Paul Kurowski
  3. Nonlinear and explicit FEA

    • Introduction to Nonlinear Finite Element Analysis, Nam-Ho Kim
    • LS-DYNA Theory Manual: explicit time integration, hourglass control, shell elements and contact
  4. Crash mechanics

    • Vehicle Crash Mechanics, Matthew Huang